According to a new report by McKinsey & Company, with more artificial intelligence and generative AI emerging, the energy demands required to power them will also increase. This energy requirement may well pose to be the new challenge for meeting net zero emissions promises that otherwise promise great benefits for a climate-friendly future. 

The report points out the increasing energy footprint of AI, with data centers—the most recent user of AI systems—appearing to be the most energy-intensive component. It predicts that power consumption by data centers in the United States will surge from the 147 TWh recorded in 2023 to 606 TWh by 2030, amounting to about 11.7% of the total power demand of the United States. While McKinsey projects the US’s emissions intensity in power generation to decline both with growth away from coal and an increase in renewables, however, it also cautions that these efforts alone are not going to be enough to meet the net-zero targets of the country. 

Despite the hype surrounding the shift to greener energy sources, McKinsey states that the sheer scale of this transition does pose a challenge to the growth of the AI industry. It chooses three areas where investors can make a difference: improving access and sources of power, advancing power equipment, and ensuring there is an adequate supply of competent technicians. 

But a contrasting Zhejiang University study paints a bleaker picture. Yet proponents of AI will tell you that increases in technological know-how lead to better efficiency and less carbon emissions, but it’s the very same study by Zhejiang University that found “12 times more energy is used to operate more sophisticated AI systems like OpenAI’s GPT-4.”. This increasing demand for computational power may worsen the environmental impact of AI, believes Meng Zhang, lead researcher at Zhejiang University and an advocate that the industry evolves to more sustainable practices. 

The biggest concern, however is the likely power supply shortages due to alarmingly fast growth of AI data centers, according to a report from Gartner. 40 percent of the AI data centers will face operational constraints because of lack of energy by 2027, and power demands might triple, says Gartner. 

The findings here reveal the need for the AI industry to incorporate into their growth all that is required globally to sustain the environment in conjunction with industrial innovations.